I've been filling out brackets for a long time. When RPI was the metric on CBS Sports, there was a comparison view that put resumes next to each other, game by game, and I loved it. They took it away, and nothing since has let me look at resumes side by side as efficiently. I've also always been fascinated by how the committee makes its decisions. I figure the better I can surface the data for myself, the closer I can get to replicating them, and seeing how close I can get as a predictive measure is the goal. Bracket Studio is where I built the version I wanted.

When I build a bracket, I pull up the top eight teams in whichever metric is primary at the time, usually NET, and compare all eight side by side. I pick the team I'd put at number one, place it on the seed list, and replace it with the team ranked ninth. Then I compare that group and pick the next one. Often I only need to weigh a couple of them. During the season a big win or loss can move a team well past the eight it's compared against, so there's a little science and a little art in it. If you want to try it, the Studio puts the same sheets on one screen, and the next bracket is yours to build.

Bracket Matrix has graded my bracket every tournament year since 2010. In 2026 it scored 374 of 408 on the Paymon scale, named all 68 teams, and put 52 of them on the exact seed line.

Every bracket here is scored with the Paymon Scoring System, which grades exact seed placement and gives partial credit for proximity. Picking the right teams and seeding them badly is a worse bracket than the selection accuracy suggests, and this score catches that.

Scores are not comparable across formats. The 2026 bracket was built for the 68-team field with a maximum of 408 points. The 76-team era raises that maximum to 456.

See the 2026 bracket scored line by line · Sixteen years, graded by Bracket Matrix

My women's bracket publishes on the forecast page too. It has no scored record yet, because no women's season of mine has been graded.